Redis further expands its presence in the machine learning software stack with the launch of Redis Feature Form, a managed feature store for production machine learning.
Following Redis’ acquisition of Featureform, this release unifies feature definition, orchestration, versioning, and servicing into a single platform. The system is designed to help machine learning teams move models from test to production while mitigating production efforts and mismatches between training data and live inference.
A recent report from Deloitte found that only a quarter of companies successfully move their artificial intelligence pilots into production. Redis targets that gap with software designed to manage the entire feature lifecycle: the structured data inputs used to train and run machine learning models.
The platform combines offline training workflows with online serving, an area that often required separate systems or custom integrations. By integrating these features, teams can define features once and use them for both model training and inference.
Platform changes
Newly added features include support for batch and streaming pipelines, including backfill, incremental updates, and tiling. Redis also introduces workspaces for multi-team use, allowing organizations to separate data, authentication, and observability by workspace.
Added job planning, impact analysis, split materialization, and queue-based job management to give teams more visibility into changes before they impact production systems. Another change is to manage graph-level updates atomically, rather than versioning individual resources individually, to simplify rollbacks and change tracking.
Security updates include workspace-level access controls, API key pairs, audit logging, secret provider updates, mTLS, and encrypted internal transport. Redis also reduced deployment complexity with two service deployment models and rebuilt the dashboard to support workspace and provider configuration through an interface.
Focus on production
Feature stores have become an important part of machine learning infrastructure because they help organizations manage data features used in development and production models. Problems arise when the data used to train the model differs from the data available at the time of deployment, which can lead to performance issues and model drift.
Redis positions the platform for production use cases such as fraud detection, credit and risk scoring, and personalization and recommendation systems, where delays or inconsistencies in feature data can have direct commercial impact. We’re also targeting platform teams that are building internal pipelines and need a more controlled path to self-service machine learning work.
This move expands Redis’ position in the market. The company has traditionally been used in large-scale machine learning architectures, particularly as a low-latency data layer for online services. Redis is seeking a greater role in defining, managing, and coordinating functionality across an organization using Feature Forms.
As a result, Redis faces increased competition from vendors that offer feature stores and broader machine learning operational tools. This also reflects the growing demand for software that can connect artificial intelligence experimental work to stable production systems without extensive custom engineering.
Simba Khadder, head of Context Engine at Redis, said the product aims to reduce complexity for machine learning teams, saying, “Feature Form helps ML teams move features from definition to production with less glue code, less drift between training and delivery, and less operational overhead.”
“It’s built for a future where Redis supports both modern AI systems and long-running ML workloads from the same data platform foundation,” Khadder said.
